Reflection - Part 3: CPOE and CDS

From Yue:

I like David mentioned that medical students are still required to remember the mnemonic "ADC VanDIMLS" although the electronic data and clinical health care system became popular nowadays. By reviewing the clinical decision support systems that were available or that are currently used in hospitals, I realize there is still a long way to make CDS trustful. Diseases are so complex that treatments should be individualized.

From Xiruo:

It is very interesting to read an academic paper discussing real practices switching from CPOE to paper-based system. The authors definitely got the points of almost all flaws that current CPOE/CDS systems could encounter and the reasons physicians keep complaining about. But just as mentioned in our main section, these issues could be roughly categorized into what we’ve already known: user interface, work flow, and stable technology. It is also amazing to find that transition paper in 2017, where the hospital made the switch in 2015 after three years of bad experiences with CPOE, and the paper about three pillars of CPOE/CDS was in 2007. It’s been ten years, or probably more, since we first realized those issues, but it seems little has been achieved in this commercial market, where softwares with flaws are still being advertised and used. Don’t we need standards for CPOE/CDS systems to make academia research be applied elsewhere? If so, who should be responsible making the standards?

From Jake:

I thought it was really interesting how doctors learn to think in terms of probability. I have a friend in medical school and I remembered getting embarrassed when he knew bayes theorem off the top of his head and I did not. Now I realize that he probably had just learned that in medical school. It's interesting to see the confusion matrix that doctors have to take into account. Doctor's are really machine learning and statistical gurus. I also learned that NLP from unstructured data is beyond where I thought it would be. NLP is also advancing in other countries too and may not translate depending on the method.